Can constructive mathematics be applied in physics? (Q1818378): Difference between revisions

Some recent algorithmic results seem to indicate, at first glance, that constructive mathematics, in its present form, may not be adequate for dealing, e.g., with Hilbert space-based mathematics behind quantum mechanics. For example, Pour-El and Richards have shown that if a linear operator \(T:H\to H\) transforms every constructive element \(h\in H\) of a Hilbert space \(H\) into a constructive element, then \(T\) has to be bounded; thus, there seems to be a problem with unbounded operators like \(d/dx\) which are necessary for quantum physics. Similarly, a seemingly natural constructive analogue of Gleason's theorem -- one of the main results of the foundations of quantum physics -- is proven to be false, which led several researchers to conclude that constructive mathematics is not a perfect foundation for quantum physics. The author argues that this conclusion is caused mainly by a misunderstanding of what constructive mathematics is about. After a brief explanation of the nature of modern constructive mathematics, the author clarifies both above examples. First, he reminds the readers that every statement from classical mathematics has several non-equivalent constructive analogues, some of them false, some true; thus, the fact that one constructive analogue of Gleason's theorem is false does not mean that other constructive analogues cannot be true -- and some actually are. Second, he analyzes Pour El-Richards results in a similar manner: a constructive analogue of the differentiation operator can be well-defined and well analyzed, and the fact that the derivative of some constructive elements \(h\in H\) is not constructive (and thus not defined in constructive mathematics) is no more surprising than the known fact that in classical mathematics the derivative is undefined for some functions \(h\in H\). The proper constructive domain of an unbounded operator \(T\) is \textit{not} the set of all constructive elements \(h\in H\) for which \(T(h)\) is classically defined but rather (a smaller set of) all constructive elements \(h\in H\) for which \(T(h)\) is \textit{constructively} defined.